Merge pull request #112 from IGNF/fix-output-epsg

Pass input file parameters to las writer

CHANGELOG.md

@@ -1,6 +1,8 @@
 # CHANGELOG
+### 3.8.1
+- fix: propagate input las format to output las (in particular epsg which comes either from input or config)
 
-### 3.8.0
+## 3.8.0
 - dev: log confusion matrices to Comet after each epoch.
 - fix: do not mix the two way to log IoUs to avoid known lightning [Common Pitfalls](https://lightning.ai/docs/torchmetrics/stable/pages/lightning.html#common-pitfalls).
 
@@ -9,7 +11,7 @@
 
 ## 3.7.0
 - Update all versions of Pytorch, Pytorch Lightning, and Pytorch Geometric.
-  Changes are retrocompatible for models trained with older versions (with adjustment to the configuration file). 
+  Changes are retrocompatible for models trained with older versions (with adjustment to the configuration file).
 - Refactor logging of single-class IoUs to go from num_classes+1 torchmetrics instances to only 1.
 
 ### 3.6.1
@@ -64,4 +66,4 @@
 - Fix dataset description for pacasam: there was an unwanted int-to-int mapping in classification_dict
 
 ## 3.4.0
-- Allow inference for the smallest possible patches (num_nodes=1) to have consistent inference behavior 
+- Allow inference for the smallest possible patches (num_nodes=1) to have consistent inference behavior

environment.yml

@@ -29,10 +29,11 @@ dependencies:
   - numpy
   - h5py
   # --------- geo --------- #
-  - pdal
+  - pdal==2.6.*
   - python-pdal
-  - pyproj
-    # --------- Visualization --------- #
+  - conda-forge:gdal
+  - conda_forge:pyproj
+  # --------- Visualization --------- #
   - pandas
   - matplotlib
   # --------- loggers --------- #
@@ -68,4 +69,5 @@ dependencies:
     # --------- Documentation --------- #
     - myst_parser==0.17.*
     - sphinxnotes-mock==1.0.0b0  # still a beta
-    - sphinx_paramlinks==0.5.*
+    - sphinx_paramlinks==0.5.*
+    - ign-pdal-tools>=1.5.2

myria3d/models/interpolation.py

@@ -8,6 +8,8 @@
 from torch.distributions import Categorical
 from torch_scatter import scatter_sum
 
+from pdaltools import las_info
+
 from myria3d.pctl.dataset.utils import get_pdal_info_metadata, get_pdal_reader
 
 log = logging.getLogger(__name__)
@@ -55,7 +57,7 @@ def __init__(
         self.logits: List[torch.Tensor] = []
         self.idx_in_full_cloud_list: List[np.ndarray] = []
 
-    def load_full_las_for_update(self, src_las: str, epsg: str) -> np.ndarray:
+    def load_full_las_for_update(self, src_las: str, epsg: str) -> Tuple[np.ndarray, Dict]:
         """Loads a LAS and adds necessary extradim.
 
         Args:
@@ -83,7 +85,10 @@ def load_full_las_for_update(self, src_las: str, epsg: str) -> np.ndarray:
             pipeline |= pdal.Filter.assign(value=f"{self.entropy_channel}=0")
 
         pipeline.execute()
-        return pipeline.arrays[0]
+        writer_params = las_info.get_writer_parameters_from_reader_metadata(
+            pipeline.metadata, a_srs=f"EPSG:{epsg}" if str(epsg).isdigit() else epsg
+        )
+        return pipeline.arrays[0], writer_params
 
     def store_predictions(self, logits, idx_in_original_cloud) -> None:
         """Keep a list of predictions made so far."""
@@ -143,7 +148,7 @@ def reduce_predictions_and_save(self, raw_path: str, output_dir: str, epsg: str)
         del logits
 
         # Read las after fetching all information to write into it
-        las = self.load_full_las_for_update(raw_path, epsg)
+        las, writer_params = self.load_full_las_for_update(raw_path, epsg)
 
         for idx, class_name in enumerate(self.classification_dict.values()):
             if class_name in self.probas_to_save:
@@ -173,9 +178,8 @@ def reduce_predictions_and_save(self, raw_path: str, output_dir: str, epsg: str)
         out_f = os.path.abspath(out_f)
         log.info(f"Updated LAS ({basename}) will be saved to: \n {output_dir}\n")
         log.info("Saving...")
-        pipeline = pdal.Writer.las(
-            filename=out_f, extra_dims="all", minor_version=4, dataformat_id=8
-        ).pipeline(las)
+        writer_params["extra_dims"] = "all"
+        pipeline = pdal.Writer.las(filename=out_f, **writer_params).pipeline(las)
         pipeline.execute()
         log.info("Saved.")
 

myria3d/pctl/dataset/utils.py

@@ -85,19 +85,12 @@ def get_pdal_reader(las_path: str, epsg: str) -> pdal.Reader.las:
 
     if epsg :
         # if an epsg in provided, force pdal to read the lidar file with it
-        try :  # epsg can be added as a number like "2154" or as a string like "EPSG:2154"
-            int(epsg)
-            return pdal.Reader.las(
-                filename=las_path,
-                nosrs=True,
-                override_srs=f"EPSG:{epsg}",
-            )
-        except ValueError:
-            return pdal.Reader.las(
-                filename=las_path,
-                nosrs=True,
-                override_srs=epsg,
-            )
+        # epsg can be added as a number like "2154" or as a string like "EPSG:2154"
+        return pdal.Reader.las(
+            filename=las_path,
+            nosrs=True,
+            override_srs=f"EPSG:{epsg}" if str(epsg).isdigit() else epsg,
+        )
 
     try :
         if get_metadata(las_path)['metadata']['readers.las']['srs']['compoundwkt']:

package_metadata.yaml

@@ -1,4 +1,4 @@
-__version__: "3.8.0"
+__version__: "3.8.1"
 __name__: "myria3d"
 __url__: "https://github.com/IGNF/myria3d"
 __description__: "Deep Learning for the Semantic Segmentation of Aerial Lidar Point Clouds"

tests/myria3d/test_train_and_predict.py

@@ -4,6 +4,8 @@
 import numpy as np
 import pytest
 from lightning.pytorch.accelerators import find_usable_cuda_devices
+from pathlib import Path
+from pdaltools import las_info
 
 
 from myria3d.pctl.dataset.toy_dataset import TOY_LAS_DATA
@@ -94,6 +96,10 @@ def test_predict_as_command(one_epoch_trained_RandLaNet_checkpoint, tmpdir):
         "task.task_name=predict",
     ]
     run_hydra_decorated_command(command)
+    output_path = Path(tmpdir) / Path(abs_path_to_toy_LAS).name
+    metadata = las_info.las_info_metadata(output_path)
+    out_pesg = las_info.get_epsg_from_header_info(metadata)
+    assert out_pesg == DEFAULT_EPSG
 
 
 def test_command_without_epsg(one_epoch_trained_RandLaNet_checkpoint, tmpdir):